loading...
running...
How to use
Pick a preset or set the source (a seafloor displacement of 4 m moves the sea surface by about that much over the source width). The wave leaves the source as two pulses; the one heading right reaches the coast, slows, steepens and runs up the beach. Watch the three gauges (G1 deep, G2 shelf, G3 coast), the run-up number and the houses that get a flood-depth mark. Drag the picture to pan, use the wheel to zoom, or press Follow wave. Space pauses, R restarts. Changing any setup slider restarts the run.
What to try. (1) Compare Thrust quake with Submarine slide: the first lowers the sea at the beach before the crest arrives. (2) Set the shelf to 40 km and the beach slope to 1:100 and read the amplification G3/G1. (3) Add a 6 m seawall and then a forest belt and compare the inundation distance. (4) Raise the tide by 2 m on a high-energy event. (5) Open the shoaling graph and see where the simulation leaves the Green's-law line (near the shore, where nonlinearity and friction take over).
The science
Why a tsunami is a shallow-water wave. An ocean 4 km deep is still "shallow" for a wave whose length is 100 km or more: the depth is far smaller than the wavelength, so the whole water column moves together. The governing equations are the nonlinear shallow-water (Saint-Venant) equations for depth h(x,t) and depth-averaged velocity u(x,t) over a bed z(x):
∂h/∂t + ∂(hu)/∂x = 0
∂(hu)/∂t + ∂(hu² + g h²/2)/∂x = -g h ∂z/∂x - g n² |u| u / h1/3
The last term is Manning friction with roughness n (about 0.025 for sea and sand, 0.03 for open land, 0.12 for a dense forest belt). The free-surface height is η = z + h.
Speed. Linear long waves travel at c = √(g h): 200 m/s (720 km/h) in 4000 m of water, 31 m/s in 100 m, 3 m/s in 1 m. This slowing is why the wave front bunches up near the coast.
Shoaling (Green's law). On a gentle slope the energy flux per metre of crest, ρ g η² c / 2 times the crest width, is conserved, so amplitude grows as η ∝ h-1/4. A 1 m open-ocean wave in 4000 m of water becomes about 2.8 m in 40 m of water, before any nonlinear steepening, and the wavelength shrinks in proportion to √h. The graph "Peak water level vs depth" compares the simulated peak with this law.
Run-up. When the wave reaches the beach the shoreline is a moving boundary: cells wet and dry, and the water climbs the slope. The run-up R is the highest land elevation reached. For a gentle beach it is larger than the offshore wave height, for a very steep wall it reflects the wave instead. Forest and wall options change the friction and the topography respectively.
Numerics. Finite-volume scheme on 2000 cells of 50 m: second-order MUSCL reconstruction (minmod limiter) of the free surface and velocity, hydrostatic reconstruction at cell faces (Audusse et al. 2004) so the lake at rest is preserved exactly and depths stay non-negative, a Rusanov flux, and a two-stage Heun time step with CFL 0.4. The left edge is open (waves leave), the right edge is a wall far inland. The four numerical checks run in this page against analytic results: lake at rest, wave speed, Green's law and mass conservation.
Honest simplifications. This is a one-dimensional transect: no spreading, refraction, diffraction or edge waves, which in nature reduce or focus energy; the displayed source heights are initial sea-surface displacements, not a fault model; dispersion (which matters for short slide-generated waves and for the trailing wave train of a real event) is ignored; turbulence, sediment, debris and building interaction are not modelled, and the town is a fixed set of marker houses that only record the maximum water depth at their site. Vertical scale is exaggerated and, in deep views, the seabed depth axis is compressed; the status line says when. Use it to understand the physics, not to assess hazard at any real place.